We present Denario, an AI multi-agent system designed to serve as a scientific research assistant. Denario can perform many different tasks, such as generating ideas, checking the literature, developing research plans, writing and executing code, making plots, and drafting and reviewing a scientific paper. The system has a modular architecture, allowing it to handle specific tasks, such as generating an idea, or carrying out end-to-end scientific analysis using Cmbagent as a deep-research backend. In this work, we describe in detail Denario and its modules, and illustrate its capabilities by presenting multiple AI-generated papers generated by it in many different scientific disciplines such as astrophysics, biology, biophysics, biomedical informatics, chemistry, material science, mathematical physics, medicine, neuroscience and planetary science. Denario also excels at combining ideas from different disciplines, and we illustrate this by showing a paper that applies methods from quantum physics and machine learning to astrophysical data. We report the evaluations performed on these papers by domain experts, who provided both numerical scores and review-like feedback. We then highlight the strengths, weaknesses, and limitations of the current system. Finally, we discuss the ethical implications of AI-driven research and reflect on how such technology relates to the philosophy of science. We publicly release the code at this https URL. A Denario demo can also be run directly on the web at this https URL, and the full app will be deployed on the cloud.

We investigate a WIMP dark matter (DM) candidate in the form of a singlino-dominated lightest supersymmetric particle (LSP) within the $Z_3$-symmetric Next-to-Minimal Supersymmetric Standard Model. This framework gives rise to regions of parameter space where DM is obtained via co-annihilation with nearby higgsino-like electroweakinos and DM direct detection~signals are suppressed, the so-called ``blind spots". On the other hand, collider signatures remain promising due to enhanced radiative decay modes of higgsinos into the singlino-dominated LSP and a photon, rather than into leptons or hadrons. This motivates searches for radiatively decaying neutralinos, however, these signals face substantial background challenges, as the decay products are typically soft due to the small mass-splits ($\Delta m$) between the LSP and the higgsino-like coannihilation partners. We apply a data-driven Machine Learning (ML) analysis that improves sensitivity to these subtle signals, offering a powerful complement to traditional search strategies to discover a new physics scenario. Using an LHC integrated luminosity of $100~\mathrm{fb}^{-1}$ at $14~\mathrm{TeV}$, the method achieves a $5\sigma$ discovery reach for higgsino masses up to $225~\mathrm{GeV}$ with $\Delta m\!\lesssim\!12~\mathrm{GeV}$, and a $2\sigma$ exclusion up to $285~\mathrm{GeV}$ with $\Delta m\!\lesssim\!20~\mathrm{GeV}$. These results highlight the power of collider searches to probe DM candidates that remain hidden from current direct detection experiments, and provide a motivation for a search by the LHC collaborations using ML methods.

Einstein-Maxwell-dilaton theory is an interesting and well-motivated theoretical laboratory to explore the impact of new fundamental degrees of freedom in the context of testing the no-hair conjecture, due to the existence of hairy black hole solutions together with the propagation of scalar, vector and tensor modes. In this paper we compute the quasinormal mode spectrum of static and slowly rotating black holes for generic values of the dilaton coupling, within a weak electric charge approximation. Our results suggest that these spacetimes are stable for generic values of the dilaton coupling and the black hole charge. We also show that while gravitational modes are only weakly affected by the coupling with the dilaton, the spectrum of electromagnetic modes exhibits a more pronounced dilaton-dependent breaking of isospectrality between the axial and polar sectors. We further show that the gravitational quasinormal modes are well approximated by the properties of unstable null circular geodesics in those spacetimes, while the treatment of electromagnetic and scalar modes can be simplified by a suitably modified Dudley-Finley scheme for the perturbed equations.

We study the dynamics of a non-minimally coupled (NMC) scalar spectator field in non-oscillatory inflationary scenarios, where there is a transition from inflation to kination domination (KD). Engineering a realistic finite-duration transition through a CMB-compatible inflaton potential, we calculate the initial tachyonic growth of the NMC field during KD and perform lattice simulations of the subsequent non-linear dynamics. We characterize the regularization effect on the tachyonic growth, either due to self-interactions, or via gravitational backreaction when the NMC field grows to dominate the energy of the universe. Our study provides the first realistic treatment of the dynamics, with significant improvements compared to previous work, where one or more of the following aspects were assumed: ($i$) the background expansion can be neglected during the tachyonic growth, ($ii$) coherence of the NMC field, ($iii$) coherence of the inflaton, ($iv$) instantaneous transition, and ($v$) a KD equation of state of exactly $w = 1$. Using our methodology, which requires none of the above assumptions, we determine the conditions to achieve proper reheating, i.e. energetic dominance of the NMC field over the inflaton. We characterize the time and energy scales of the problem, either for backreaction due to self-interactions, or (as a novelty of this work) due to gravitational effects. Finally, we calculate $\mathcal{O}(1)$ lattice correction factors to analytic scaling relations derived by some of us in previous work. This enables simple future studies without the need to run lattice simulations.

This letter explores a transition in the type of von Neumann algebra for asymptotically AdS spacetimes from the implementations of the different gravitational constraints. We denote it as the \emph{centaur-algebra} of observables. In the first part of the letter, we employ a class of flow geometries interpolating between AdS$_2$ and dS$_2$ spaces, the centaur geometries. We study the type II$_\infty$ crossed product algebra describing the semiclassical gravitational theory, and we explore the algebra of bounded sub-regions in the bulk theory following $T\overline{T}$ deformations of the geometry and study the gravitational constraints with respect to the quasi-local Brown-York energy of the system at a finite cutoff. In the second part, we study arbitrary asymptotically AdS spacetimes, where we implement the boundary protocol of an infalling observer modeled as a probe black hole proposed by arXiv:2211.16512 to study modifications in the algebra. In both situations, we show how incorporating the constraints requires a type II$_1$ description.

The continued development of computational approaches to many-body ground-state problems in physics and chemistry calls for a consistent way to assess its overall progress. In this work, we introduce a metric of variational accuracy, the V-score, obtained from the variational energy and its variance. We provide an extensive curated dataset of variational calculations of many-body quantum systems, identifying cases where state-of-the-art numerical approaches show limited accuracy, and future algorithms or computational platforms, such as quantum computing, could provide improved accuracy. The V-score can be used as a metric to assess the progress of quantum variational methods toward a quantum advantage for ground-state problems, especially in regimes where classical verifiability is impossible.

We complete the classification of Mink$_4$ solutions preserving $\mathcal{N}=2$ supersymmetry and SU(2) R-symmetry parameterised by a round $S^2$ factor. We consider eleven-dimensional supergravity and relax the assumptions of earlier works in type II theories. We show that, using chains of dualities, all solutions of this type can be generated from one of two master classes: an SU(2)-structure in M-theory and a conformal Calabi-Yau in type IIB. Finally, using our results, we recover AdS$_5\times S^2$ solutions in M-theory and construct a compact Minkowski solution with Atiyah-Hitchin singularity.

We introduce SpectraPyle, a versatile spectral stacking pipeline developed for the Euclid mission's NISP spectroscopic surveys, aimed at extracting faint emission lines and spectral features from large galaxy samples in the Wide and Deep Surveys. Designed for computational efficiency and flexible configuration, SpectraPyle supports the processing of extensive datasets critical to Euclid's non-cosmological science goals. We validate the pipeline using simulated spectra processed to match Euclid's expected final data quality. Stacking enables robust recovery of key emission lines, including Halpha, Hbeta, [O III], and [N II], below individual detection limits. However, the measurement of galaxy properties such as star formation rate, dust attenuation, and gas-phase metallicity are biased at stellar mass below log10(M*/Msol) ~ 9 due to the flux-limited nature of Euclid spectroscopic samples, which cannot be overcome by stacking. The SFR-stellar mass relation of the parent sample is recovered reliably only in the Deep survey for log10(M*/Msol) > 10, whereas the metallicity-mass relation is recovered more accurately over a wider mass range. These limitations are caused by the increased fraction of redshift measurement errors at lower masses and fluxes. We examine the impact of residual redshift contaminants that arises from misidentified emission lines and noise spikes, on stacked spectra. Even after stringent quality selections, low-level contamination (< 6%) has minimal impact on line fluxes due to the systematically weaker emission of contaminants. Percentile-based analysis of stacked spectra provides a sensitive diagnostic for detecting contamination via coherent spurious features at characteristic wavelengths. While our simulations include most instrumental effects, real Euclid data will require further refinement of contamination mitigation strategies.

This work develops a rigorous numerical framework for solving time-dependent Optimal Control Problems (OCPs) governed by partial differential equations, with a particular focus on biomedical applications. The approach deals with adjoint-based Lagrangian methodology, which enables efficient gradient computation and systematic derivation of optimality conditions for both distributed and concentrated control formulations. The proposed framework is first verified using a time-dependent advection-diffusion problem endowed with a manufactured solution to assess accuracy and convergence properties. Subsequently, two representative applications involving drug delivery are investigated: (i) a light-triggered drug delivery system for targeted cancer therapy and (ii) a catheter-based drug delivery system in a patient-specific coronary artery. Numerical experiments not only demonstrate the accuracy of the approach, but also its flexibility and robustness in handling complex geometries, heterogeneous parameters, and realistic boundary conditions, highlighting its potential for the optimal design and control of complex biomedical systems.

A systematically improvable wave function is proposed for the numerical solution of strongly correlated systems. With a stochastic optimization method, based on the auxiliary field quantum Monte Carlo technique, an effective temperature Teff is defined, probing the distance of the ground state properties of the model in the thermodynamic limit from the ones of the proposed correlated mean-field ansatz. In this way their uncertainties from the unbiased zero temperature limit may be estimated by simple and stable extrapolations well before the so called sign problem gets prohibitive. At finite Teff the convergence of the energy to the thermodynamic limit is indeed shown to be possible in the Hubbard model already for relatively small square lattices with linear dimension L ~10, thanks to appropriate averages over several twisted boundary conditions. Within the estimated energy accuracy of the proposed variational ansatz, two clear phases are identified, as the energy is lowered by spontaneously breaking some symmetries satisfied by the Hubbard Hamiltonian: a) a stripe phase where both spin and translation symmetries are broken, and b) a strong coupling d-wave superconducting phase when the particle number is not conserved and global U(1) symmetry is broken. On the other hand the symmetric phase is stable in a wide region at large doping and small coupling.

The recent observation of the $K^+ \to \pi^+ \nu\bar\nu$ decay by NA62 is an important milestone in precision flavor physics. Together with evidence of $B^+ \to K^+\nu\bar\nu$ reported by Belle-II, they are the only FCNC decays involving third-family leptons where a precision close to the SM expectation has been reached. We study the implications of these recent results in the context of a new physics scenario aligned to the third generation, with an approximate $U(2)^5$ flavor symmetry acting on the light families. We find that the slight excess observed in both channels supports the hypothesis of non-standard TeV dynamics of this type, as also hinted at by other $B$-meson decays, consistently with bounds from colliders and electroweak observables. We further discuss how future improvements in precision could affect this picture, highlighting the discovery potential in these di-neutrino modes.

There has been recently considerable progress in understanding the nature of perturbation theory in UV free and gapped $2d$ integrable field theories with renormalon singularities. Thanks to Bethe ansatz and large $N$ techniques, non-perturbative corrections can also be computed and lead to the reconstruction of the trans-series for the free energy in presence of a chemical potential. This is an ideal arena to test resurgence in QFT and determine if and how the exact result can be reconstructed from the knowledge of the perturbative series only. In these notes we give a pedagogical introduction to this subject starting from the basics. In the first lecture we give an overview of applications in QFT of Borel resummations before the advent of resurgence. The second lecture introduces the key concepts of resurgence and finally in the third lecture we discuss a specific application in the context of the principal chiral field model. Extended version of three lectures given at IHES and review talks given at Les Diablerets and Mainz, in 2023.

We study the stochastic dynamics of a symmetric self-chemotactic particle and determine the long-time behavior of its mean squared displacement (MSD). The attractive or repulsive interaction of the particle with the chemical field that it generates induces a non-linear, non-Markovian effective dynamics, which results into anomalous diffusion for spatial dimensions $d \leq 2$. In one spatial dimension, we map the case of repulsive chemotaxis onto a run-and-tumble-like dynamics, leading to an MSD which, as a function of the elapsed time $t$, grows superdiffusively with exponent $4/3$. In the presence of attractive chemotaxis, instead, the particle exhibits a slowdown, with the MSD growing logarithmically with time. In $d=2$, we find logarithmic aging of the diffusion coefficient, while in $d=3$ the motion reverts standard diffusive behavior with a renormalized diffusion coefficient.

Modifications of Einstein's theory of gravitation have been extensively considered in the past years, in connection to both cosmology and quantum gravity. Higher-curvature and higher-derivative gravity theories constitute the main examples of such modifications. These theories exhibit, in general, more degrees of freedom than those found in standard General Relativity; counting, identifying, and retrieving the description/representation of such dynamical variables is currently an open problem, and a decidedly nontrivial one. In this work we review, via both formal arguments and custom-made examples, the most relevant methods to unveil the gravitational degrees of freedom of a given model, discussing the merits, subtleties and pitfalls of the various approaches.

The ultralight dark photon is a well-motivated, hypothetical dark matter candidate. In a dilute plasma, they can resonantly convert into photons, and heat up the intergalactic medium between galaxies. In this work, we explore the dark photon dark matter parameter space by comparing synthetic Lyman-$\alpha$ forest data from cosmological hydrodynamical simulations to observational data from VLT/UVES of the quasar HE0940-1050 ($z_{\rm em}=3.09$). We use a novel flux normalization technique that targets under-dense gas, reshaping the flux probability distribution. Not only do we place robust constraints on the kinetic mixing parameter of dark photon dark matter, but notably our findings suggest that this model can still reconcile simulated and observed Doppler parameter distributions of $z\sim0$ Lyman-$\alpha$ lines, as seen by HST/COS. This work opens new pathways for the use of the Lyman-$\alpha$ forest to explore new physics, and can be extended to other scenarios such as primordial black hole evaporation, dark matter decay, and annihilation.

Through well-motivated models in particle physics, we demonstrate the power of a general class of selection rules arising from non-invertible fusion algebras that are only exact at low orders in perturbation theory. Surprisingly, these non-invertible selection rules can even be applied to the minimal extension of the Standard Model, which is to add a gauge-singlet real scalar. In this model, we show that Fibonacci fusion rules lead to experimentally testable features for the scattering processes of the real scalar. We anticipate that this class of non-invertible selection rules can be applied to a wide range of models beyond the Standard Model. To further strengthen our methodology, we discuss a dark matter model based on the Ising fusion rules, where the dark matter is labeled by the non-invertible element in the algebra, hence its stability is preserved at all loop orders.

In the presence of a magnetic field, axions can convert into photons and vice versa. The phenomenology of the conversion is captured by a system of two coupled Klein-Gordon equations, which, assuming that the axion is relativistic, is usually recast into a pair of first-order Schrödinger-like equations. In such a limit, focusing on a constant magnetic field and plasma frequency, the equations admit an exact analytic solution. The relativistic limit significantly simplifies the calculations and, therefore, it is widely used in phenomenological applications. In this work, we discuss how to evaluate the axion-photon system evolution without relying on such relativistic approximation. In particular, we give an exact analytical solution, valid for any axion energy, in the case that both the magnetic field and plasma frequency are constant. Moreover, we devise an analytic perturbative expansion that allows for tracking the conversion probability in a slightly inhomogeneous magnetic field or plasma frequency, whose characteristic scale of variation is much larger than the typical axion-photon oscillation length. Finally, we discuss a case of resonant axion-photon conversion giving useful simplified formulae that might be directly applied to dark matter axions converting in neutron star magnetospheres.

Abstraction is the process of extracting the essential features from raw data while ignoring irrelevant details. This is similar to the process of focusing on large-scale properties, systematically removing irrelevant small-scale details, implemented in the renormalisation group of statistical physics. This analogy is suggestive because the fixed points of the renormalisation group offer an ideal candidate of a truly abstract -- i.e. data independent -- representation. It has been observed that abstraction emerges with depth in neural networks. Deep layers of neural network capture abstract characteristics of data, such as "cat-ness" or "dog-ness" in images, by combining the lower level features encoded in shallow layers (e.g. edges). Yet we argue that depth alone is not enough to develop truly abstract representations. We advocate that the level of abstraction crucially depends on how broad the training set is. We address the issue within a renormalisation group approach where a representation is expanded to encompass a broader set of data. We take the unique fixed point of this transformation -- the Hierarchical Feature Model -- as a candidate for an abstract representation. This theoretical picture is tested in numerical experiments based on Deep Belief Networks trained on data of different breadth. These show that representations in deep layers of neural networks approach the Hierarchical Feature Model as the data gets broader, in agreement with theoretical predictions.